Bloomington (area) is host to NetSci 2006, a new international workshop/conference on Network Science. This week has 5 days and 16 sessions worth of 2-hour lectures on all things network with some of this domain’s movers and shakers in attendance.
James Moody (Ohio State) led things off with a much appreciated overview of social networking. Lots of terms that proved helpful the rest of the day. I took lots of notes, but mostly I tried to digest it all with an eye toward relevancy with Stone Center’s concept of mutuality. The topic was on diffusion of goods over a dynamic network, first covering the basics of a static network and then showing how having the network change over time affects these calculations. This is apparently a wide-open area of research that has great implications on my desire to build online forums that increase and sustain empathic relationships. Mostly, I compiled a long list of questions to answer through research (either mine or someone else’s work). The most interesting were:
- When does a relationship end?
- How does mutuality increase concurrency and the ability to make new connections through reference and proximity
Lada Adamic (Michigan) came next with a nice talk about searches in networks. Again, a brief overview of experiments and results of small-world networks was helpful. Mostly, this identified geography, profession and hierarchy as the three most utilized strategies for propagating or locating information. Lada then talked about four research studies:
- HP Labs Email — examined search strategies of her former employer when looking for information via email. Use of informal hierarchies proved most efficient.
- Club Nexus — Looked at 2500 users of an online community with self-reported profile information. The results showed this network was barely searchable.
- Killworth/McCarty/Bernard/House — This not-yet-published study in asked people to list their social network and then identify the top contact. Just over half of the messages would have reached a target destination using this local information, but that also meant that half of all decisions proved ineffective.
- LiveJournal — 1 million active bloggers in a network of about 10 million use this tool. This study looked to find a target’s city. It turns out the distance is not as important as the density of people in between a node and the target.
The take-home for me here was that, like time, the value of a relationship and the number of nodes in close proximity changes the definition of “shortest path.” I’d like to know how mutuality impacts these things, in the hope of identifying obstacles or catalysts to make stronger networks.
The afternoon was not as fruitful. After a nice lunch with surprisingly good offerings for a vegetarian, it was two hours of math with Carter Butts. He’s a good speaker, actually, and started out with a very appealing topic title — “Relation Associations” — but the content was primarily mechanical in nature and lost me for good with about 30-40 minutes to go. In the poster session, I primarily wanted to see one exhibit (“Networkbased Ranking System for U.S. College Football“) but Mark Newman didn’t show up yet, so my sports informatics-y draw never materialized. Newman presents on Wed. morning, though, so maybe I’ll try to ask some questions then.
Tomorrow, our own Alex Vespignani presents first, followed by Newman, Zoltán Toroczkai and Jeff Johnson. Great lineup, although I hope I have enough stamina to get to Johnson’s lecture on social capital at the end.
Check out the NetSci Blog for more coverage and links to papers and presentation files.
